To filter out noise from a voice that is transferred by a mobile phone to a peer user, most mobile phones have multiple microphones disposed, and multiple microphone input signals collected by the multiple microphones are processed by means of multi-microphone noise-canceling to obtain an uplink output signal. The multi-microphone noise-canceling is generally forming a sound pickup range having a specific geometric shape near a mobile phone, collecting an uplink user voice within the sound pickup range, and suppressing ambient noise out of the sound pickup range. For example, arrangement of a typical two-microphone solution is that one microphone is disposed near a top of a mobile phone, and is mainly configured to collect ambient noise. The other microphone is disposed at a bottom of the mobile phone, and collects a user voice including ambient noise. The mobile phone processes two microphone input signals by means of multi-microphone noise-canceling, eliminates the ambient noise in the user voice including ambient noise to obtain a relatively clear user voice, and transfers the user voice to a peer end.
From a perspective of whether a sound pickup range is fixed, multi-microphone noise-canceling may be classified into two types. An effect of a fixed sound pickup range is relatively stable. However, when a user's hand posture is abnormal, for example, when a mobile phone deviates from a user's face by a large angle, uplink silence, little sound, or poor quality may occur because a user's mouth deviates from the fixed sound pickup range. If a relatively large sound pickup range is merely set, an effect of filtering ambient noise is lessened. To better balance an uplink noise-canceling effect and adaptability of a user's hand posture, there are some multi-microphone noise-canceling solutions at present. In the solutions, a voice call frequency-band signal collected by a microphone is used to estimate a user's hand posture, and a multi-microphone noise-canceling sound pickup range of a mobile phone is controlled according to the estimated user's hand posture. In this way, ambient noise is filtered out, as much as possible, using a relatively small sound pickup range when a user's hand posture is normal, and obvious adverse impact on volume or voice quality of an uplink user voice is avoided by increasing a sound pickup range when a hand posture is abnormal.
In a process of implementing the present disclosure, the inventor finds that other approaches have at least the following problem.
Ambient noise easily affects estimation of a user's hand posture using a voice frequency-band signal. Consequently, accuracy of controlling a multi-microphone noise-canceling sound pickup range is low.